More feature creepiness - dimensional obfuscation and multi-gigapixel digital images

In a post that appeared earlier today, I said the following:

There are numerous examples of ways in which feature creep results in us buying features that we never use. How many people use their sport utility vehicles (SUVs) for...well, for sport utility?

In that earlier post, I failed to mention the one product that REALLY exhibits feature creepiness.

Digital cameras.

Before I talk about today's digital cameras, let's go back a few years to 2003 and see what the New York Institute of Photography was recommending for digital camera resolution.

At this point, we recommend purchasing a camera that is capable of capturing images of three million pixels (that's three megapixels) or more. There are lots of situations when you'll want to capture smaller files, but if you want to print a good "8x10" image you'll need a minimum of three megapixels. Four megapixels might be even better if you find a camera that's in your price range.

Again, this was written in 2003, when you had to pay a pretty penny to get four megapixel resolution on your camera. Fast-forward eight years, and you still had to lay out a pretty penny to get a high-resolution camera - but the scale had changed a bit.

The new top-end model from medium-format camera maker Hasselblad is now on the market, and it's not cheap: the 200-megapixel H4D-200MS will set you back 32,000 euros, or about $45,000.

If you've decided to hold off on that BMW purchase and buy a camera instead, go here for information.

But hey, isn't that 200 megapixel camera a hundred times better than a 2 megapixel camera? According to Ken Rockwell, no.

For normal 4x6" (10x15cm) prints, even VGA (640 x 480 or 0.3MP) resolution is just fine. Digital cameras did this back in 1991!

In 1999 when digital cameras were only 1.2 or 2 MP, each megapixel mattered if you were making bigger prints.

Today, even the cheapest cameras have at least 5 or 6 MP, which enough for any size print. How? Simple: when you print three-feet (1m) wide, you stand further back. Print a billboard, and you stand 100 feet back. 6MP is plenty.

Sharpness depends more on your photographic skill than the number of megapixels, because most people's sloppy technique or subject motion blurs the image more than the width of a microscopic pixel.

And there's a mathematical basis for Rockwell's assertion:

Pixel Count, expressed as Megapixels, is simply multiplying the number of horizontal pixels by the number of vertical pixels. It's exactly like calculating area. A 3 MP camera has 2,048 (horizontal) x 1,536 (vertical) pixels, or 3,145,728 pixels. We call this simply 3 MP....

It only takes a 40% increase in linear dimensions to double the pixel count! Doubling pixel count only increases the real, linear resolution by 40%, which is pretty much invisible.

"But," you may be saying to yourself, "there are multi-GIGApixel images floating around. And that's like cool and stuff."

Let's take a look at one such image, the 150 gigapixel image of Tokyo.

Over the course of six hours, photographer and 360cities.net founder Jeffery Martin captured the fascinating 600,000 pixel-wide panorama upon the roof of the lower observation deck on the Tokyo Tower. A high-powered Fujitsu computer -- packing 192GB of RAM, 2 quad-core Xeon processors, and a 4GB graphics card -- spent three months stitching together the mosaic from 10,000 individual photos captured by Martin.

Yes, that 150 gigapixel image consists of 10,000 mere mortal images (OK, they're still pretty high resolution, but not by much). Interestingly enough, the $45,000 camera does something similar, but it only combines six images into a single image.

So we only have to wait a few months, or perhaps a few weeks, for someone to create a TERApixel image to make everyone drool. Just combine 100,000 10 megapixel images and you've got it. Or spend $45,000 for the 200 megapixel camera, stitch a mere 5,000 pixels together, and you have your terapixel panorama.

Meanwhile, most of us will still take 640 x 480 pictures and be happy.